news_popularity_prediction

For news popularity prediction

Dependency:

• python>=3.6
• torch>=1.6.0
• dgl>=0.6.1
• jieba>=0.42.1

Data

The data is available at https://pan.baidu.com/s/13zmxj7o-oRRQXFcVOOiNYw, The extraction code is haju.

The code of F4 model we proposed consists of two parts. data process and model train.

data process

we use the code in data_process folder run 1_disa_csvreader.py, 2_vocab_news_creat.py, 3_entitiy_named.py, 4_creat_dataset.py, 5_calw2sTFIDF.py in Chinese datasets. run 1_mind_csvreader.py, 2_vocab_news_creat.py, 3_entitiy_named.py, 4_creat_dataset.py, 5_calw2sTFIDF.py in MIND datasets.

train

python fhgnntrain.py

test

testfhgnn.py